Html 如何在FabricJS中创建更强的模糊？

我正在使用FabricJS将图像上传到画布上，我想对其进行模糊处理。看看他们的例子，我可以创建一个光模糊。然而，我想让这个模糊更强烈。基本上，我想使这个更强大的模糊，然后有文字在上面。我想我需要调整卷积矩阵，但我不确定该怎么做

有什么提示吗

---

另外，我会上传一个提琴，但由于画布和CORS图像的问题，它不会有用：（

我解决这个问题的方法是使过滤器更大、更密集（我不完全确定它是如何工作的，但它可以工作，未来仍在寻找ELI5）

框模糊输出卷积矩阵内像素值的平均值。方法是创建一个大小为NxN的卷积矩阵，其中每个权重为1/（NxN）。这样，矩阵内的每个像素对输出图像的贡献相等，权重之和为1

矩阵

    // weak convolution filter box blur
    [ 1/9, 1/9, 1/9,
      1/9, 1/9, 1/9,
      1/9, 1/9, 1/9 ]

    // stronger convolution filter box blur
    [ 1/18, 1/18, 1/18,
      1/18, 1/18, 1/18,
      1/18, 1/18, 1/18,
      1/18, 1/18, 1/18,
      1/18, 1/18, 1/18,
      1/18, 1/18, 1/18 ]

    // strong convolution filter box blur
    [ 1/27, 1/27, 1/27,
      1/27, 1/27, 1/27,
      1/27, 1/27, 1/27,
      1/27, 1/27, 1/27,
      1/27, 1/27, 1/27,
      1/27, 1/27, 1/27,
      1/27, 1/27, 1/27,
      1/27, 1/27, 1/27,
      1/27, 1/27, 1/27 ]

    // super strong convolution filter box blur
    [ 1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36 ]

输出

矩阵

    // weak convolution filter box blur
    [ 1/9, 1/9, 1/9,
      1/9, 1/9, 1/9,
      1/9, 1/9, 1/9 ]

    // stronger convolution filter box blur
    [ 1/18, 1/18, 1/18,
      1/18, 1/18, 1/18,
      1/18, 1/18, 1/18,
      1/18, 1/18, 1/18,
      1/18, 1/18, 1/18,
      1/18, 1/18, 1/18 ]

    // strong convolution filter box blur
    [ 1/27, 1/27, 1/27,
      1/27, 1/27, 1/27,
      1/27, 1/27, 1/27,
      1/27, 1/27, 1/27,
      1/27, 1/27, 1/27,
      1/27, 1/27, 1/27,
      1/27, 1/27, 1/27,
      1/27, 1/27, 1/27,
      1/27, 1/27, 1/27 ]

    // super strong convolution filter box blur
    [ 1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36 ]

输出

矩阵

    // weak convolution filter box blur
    [ 1/9, 1/9, 1/9,
      1/9, 1/9, 1/9,
      1/9, 1/9, 1/9 ]

    // stronger convolution filter box blur
    [ 1/18, 1/18, 1/18,
      1/18, 1/18, 1/18,
      1/18, 1/18, 1/18,
      1/18, 1/18, 1/18,
      1/18, 1/18, 1/18,
      1/18, 1/18, 1/18 ]

    // strong convolution filter box blur
    [ 1/27, 1/27, 1/27,
      1/27, 1/27, 1/27,
      1/27, 1/27, 1/27,
      1/27, 1/27, 1/27,
      1/27, 1/27, 1/27,
      1/27, 1/27, 1/27,
      1/27, 1/27, 1/27,
      1/27, 1/27, 1/27,
      1/27, 1/27, 1/27 ]

    // super strong convolution filter box blur
    [ 1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36 ]

输出

矩阵

    // weak convolution filter box blur
    [ 1/9, 1/9, 1/9,
      1/9, 1/9, 1/9,
      1/9, 1/9, 1/9 ]

    // stronger convolution filter box blur
    [ 1/18, 1/18, 1/18,
      1/18, 1/18, 1/18,
      1/18, 1/18, 1/18,
      1/18, 1/18, 1/18,
      1/18, 1/18, 1/18,
      1/18, 1/18, 1/18 ]

    // strong convolution filter box blur
    [ 1/27, 1/27, 1/27,
      1/27, 1/27, 1/27,
      1/27, 1/27, 1/27,
      1/27, 1/27, 1/27,
      1/27, 1/27, 1/27,
      1/27, 1/27, 1/27,
      1/27, 1/27, 1/27,
      1/27, 1/27, 1/27,
      1/27, 1/27, 1/27 ]

    // super strong convolution filter box blur
    [ 1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36 ]

输出

---

我解决这个问题的方法是使过滤器更大、更密集（我不完全确定它是如何工作的，但它确实起到了作用，未来仍在寻找ELI5）

框模糊输出卷积矩阵内像素值的平均值。方法是创建一个大小为NxN的卷积矩阵，其中每个权重为1/（NxN）。这样，矩阵内的每个像素对输出图像的贡献相等，权重之和为1

矩阵

    // weak convolution filter box blur
    [ 1/9, 1/9, 1/9,
      1/9, 1/9, 1/9,
      1/9, 1/9, 1/9 ]

    // stronger convolution filter box blur
    [ 1/18, 1/18, 1/18,
      1/18, 1/18, 1/18,
      1/18, 1/18, 1/18,
      1/18, 1/18, 1/18,
      1/18, 1/18, 1/18,
      1/18, 1/18, 1/18 ]

    // strong convolution filter box blur
    [ 1/27, 1/27, 1/27,
      1/27, 1/27, 1/27,
      1/27, 1/27, 1/27,
      1/27, 1/27, 1/27,
      1/27, 1/27, 1/27,
      1/27, 1/27, 1/27,
      1/27, 1/27, 1/27,
      1/27, 1/27, 1/27,
      1/27, 1/27, 1/27 ]

    // super strong convolution filter box blur
    [ 1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36 ]

输出

矩阵

    // weak convolution filter box blur
    [ 1/9, 1/9, 1/9,
      1/9, 1/9, 1/9,
      1/9, 1/9, 1/9 ]

    // stronger convolution filter box blur
    [ 1/18, 1/18, 1/18,
      1/18, 1/18, 1/18,
      1/18, 1/18, 1/18,
      1/18, 1/18, 1/18,
      1/18, 1/18, 1/18,
      1/18, 1/18, 1/18 ]

    // strong convolution filter box blur
    [ 1/27, 1/27, 1/27,
      1/27, 1/27, 1/27,
      1/27, 1/27, 1/27,
      1/27, 1/27, 1/27,
      1/27, 1/27, 1/27,
      1/27, 1/27, 1/27,
      1/27, 1/27, 1/27,
      1/27, 1/27, 1/27,
      1/27, 1/27, 1/27 ]

    // super strong convolution filter box blur
    [ 1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36 ]

输出

矩阵

    // weak convolution filter box blur
    [ 1/9, 1/9, 1/9,
      1/9, 1/9, 1/9,
      1/9, 1/9, 1/9 ]

    // stronger convolution filter box blur
    [ 1/18, 1/18, 1/18,
      1/18, 1/18, 1/18,
      1/18, 1/18, 1/18,
      1/18, 1/18, 1/18,
      1/18, 1/18, 1/18,
      1/18, 1/18, 1/18 ]

    // strong convolution filter box blur
    [ 1/27, 1/27, 1/27,
      1/27, 1/27, 1/27,
      1/27, 1/27, 1/27,
      1/27, 1/27, 1/27,
      1/27, 1/27, 1/27,
      1/27, 1/27, 1/27,
      1/27, 1/27, 1/27,
      1/27, 1/27, 1/27,
      1/27, 1/27, 1/27 ]

    // super strong convolution filter box blur
    [ 1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36 ]

输出

矩阵

    // weak convolution filter box blur
    [ 1/9, 1/9, 1/9,
      1/9, 1/9, 1/9,
      1/9, 1/9, 1/9 ]

    // stronger convolution filter box blur
    [ 1/18, 1/18, 1/18,
      1/18, 1/18, 1/18,
      1/18, 1/18, 1/18,
      1/18, 1/18, 1/18,
      1/18, 1/18, 1/18,
      1/18, 1/18, 1/18 ]

    // strong convolution filter box blur
    [ 1/27, 1/27, 1/27,
      1/27, 1/27, 1/27,
      1/27, 1/27, 1/27,
      1/27, 1/27, 1/27,
      1/27, 1/27, 1/27,
      1/27, 1/27, 1/27,
      1/27, 1/27, 1/27,
      1/27, 1/27, 1/27,
      1/27, 1/27, 1/27 ]

    // super strong convolution filter box blur
    [ 1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36,
      1/36, 1/36, 1/36 ]

输出

---

看看我的要点：

---

相同的修复现在应用于要点。

在这里查看我的要点：

---

相同的修复现在应用于要点。

可能多次应用模糊以创建更强的模糊？好主意，但问题是它需要很长时间：（.我尝试过：p可能多次应用模糊以创建更强的模糊？好主意，但问题是它需要很长时间：（.我试过了：p这太棒了，兄弟..比你灵感来源的代码更好。谢谢你这太棒了，兄弟..比你灵感来源的代码更好。谢谢